Research on Intelligent Dispatching Algorithm of Mobile Energy Storage Equipment Based on Deep Learning

In this paper, an intelligent scheduling model of mobile energy storage equipment based on deep learning is proposed, which combines graph-convolutional network (GCN), Q-learning and genetic algorithm (GA) to realize dynamic perception and global optimization of complex power grid environment. Firstly, the grid topology is modeled by GCN, and the spatial characteristics of the grid are extracted and the interaction between devices is captured, which provides a global perspective for the dispatching strategy. Secondly, the scheduling strategy is constructed by Q-learning, and the value function of the state-action pair is updated by interacting with the environment, and the optimal strategy is gradually approached. In order to overcome the slow convergence speed and local optimization of Q-learning, GA is introduced to optimize the exploration mechanism, and the diversity and convergence efficiency of strategies are improved through global search. The experimental results show that the model in this paper has excellent performance in dynamic power grid, strong robustness and real-time optimization ability, and is suitable for complex dynamic environment. This paper provides an innovative solution to the scheduling problem of mobile energy storage equipment, and provides a new idea for the combination of deep learning and optimization algorithm to be applied to the resource allocation problem of complex systems.